Should the government set up a separate ministry for all environmental issues, or ask all the existing decentralized and regional political organizations to have an environment mandate? This paper studies how organizational forms, i.e. the “shell” of the bureaucracy, incentivize political agents to internalize environment related externalities. The theory of M-form versus U-form organizations features the spillovers among different bureaucratic tasks, as well as the spatial environment externality that one region (e.g.upstream) generates to another (e.g. downstream). I show that the optimality of organizational form depends crucially on the intensity of these externalities, and M-form reform might only work to deliver better environmental performances when the spatial pollution externality is less salient and the negative cross task spillovers are stronger. I further argue that yardstick competition among bureaucrats, which is often believed to favor M-form organization, may exacerbate the environmental problem. I use the empirical setting of River Chief System in China, a shift from U-form to M-form in river
management, to test the theoretical implications. With a city level dataset of daily water quality observations and environmental management information from 2015 to 2020, the empirical results emphasize the heterogeneity of the effectiveness of M-form reform on river quality, which are largely consistent with the theoretical predictions.